Open swj0418 opened 1 year ago
We do not do experiments on truncated latent code since we focus on W space instead of W-plus space. I guess the truncation trick barely influences the results since we have used $w_{avg}$ to compute the direction, and it still works.
Hello,
I could not find any reports on how the truncation trick might affects the manipulation quality.
In your code, I see that you are repeating the W code by the number of style transformations, but no further operation on WP code. If you have previously experimented with truncation trick, can you share any recollection of the results?
Thanks